standard deviation within each replicate. When this was not possible due to lack of replication 
(e.g. flow cytometry), it was assumed that the coefficient of variation was constant across all 
data values. The mean measurement (and standard deviation) for each sample for each analytic 
method is shown in Tables C.l and C.2 for > 50 pm and 10-50 pm samples respectively. 
The various CFA devices were also compared to each other and to microscopy counts 
using the raw output of the variable fluorescence (Fv), with no manipulation or transformation 
of the raw data signals, to eliminate variation caused by device-specific conversion factors. This 
F v was measured as the difference of chlorophyll fluorescence of dark-adapted phytoplankton 
(F 0 ), and the maximal fluorescence (F m ) under saturating light. As a result, the total active 
chlorophyll fluorescence (Fv) in the subsample is F v =F m -F 0 (Wright et al., 2015). These 
comparisons include the Pearson correlation coefficient (r) to quantify the strength of the 
relationship between each pair, r is useful to describe the total variability in the relationship 
between two methods even when methods are not measured in the same units, but note that 
it is dependent on the number of data points and the data range (Stockl et al., 1998), which 
limits comparability. 
Importantly, variation in all pair-wise relationships explored is not only attributed to 
analytical differences and/or imprecision, but also sample-related effects (e.g. variation in the 
true number of individuals in each subsample)(Miller et al., 2011). 
3.0 Results 
Natural variation in marine plankton communities throughout the voyage led to high 
variability in the concentration and composition of organisms observed in each size class across